Immunotoxicant Screening and Prioritization in the Twenty-first Century

Abstract

Current immunotoxicity testing guidance for drugs, high production–volume chemicals, and pesticides specifies the use of animal models to assess potential biomarkers of immune system effects (e.g., lymphoid organ and bone marrow indices, histopathology) or actual measures of immune function (e.g., responses to challenge with antigens or pathogens). These assays are resource intensive and often require special training or experience to ensure reliable results. Alternative in vitro assays to detect immunosuppression and allergic hypersensitivity have the potential to reduce animal use and testing costs and increase immunotoxicity screening and prioritization efforts. Alternative models to detect immunosuppression tend to address broad modes of action because suppression may be caused by a wide variety of events; current in vitro models access the supply of innate and adaptive immune system cells as well as cellular markers associated with function, including gene expression, protein synthesis, and proliferation. Events leading to the induction of allergic hypersensitivity, particularly contact hypersensitivity, are more restricted, and alternative methods currently exploit chemical properties and activation of defined cell populations to detect and estimate the potency of skin sensitizers.

Introduction

Current immunotoxicity hazard identification and risk assessment practices rely exclusively on animal data to detect reduced or inappropriately modulated immune system function, in the case of immunosuppression, or evidence of allergic sensitization, in the case of hypersensitivity. Some routine toxicity testing schemes that are not designed to specifically assess effects on the immune system (e.g., the OECD 407 repeated-dose oral toxicity guidance) do assess lymphoid organs, hematology, and bone marrow, thus providing screening data, but negative data are not conclusive and positive data typically trigger additional testing to determine whether immune function is suppressed. The U.S. Environmental Protection Agency (EPA) now requires evaluation of immune function (the antibody response to a novel antigen) for pesticide registration (US EPA 2007) rather than screening using data from other required tests. Allergic hypersensitivity, particularly contact hypersensitivity, is relatively common in the general population and is associated with exposure to industrial chemicals, personal care products, and drugs. As such, animal testing is often required (e.g., EPA OPPTS 870.2600 for pesticides, US EPA 2003). Originally, testing was done in guinea pigs (OECD 406), required ten or twenty animals per group, and spanned thirty days of exposure and assessment. An update guideline (OECD 429) describes alternative testing methods in smaller groups of mice (the local lymph node assay; LLNA), spanning five days of exposure and testing. Although this method meets the goal of reduced animal use and faster assessment, it is far from medium or high throughput.

There is growing political and practical resistance to continued reliance on traditional testing methods to screen industrial and commercial chemicals and consumer products, which is driving the development of alternatives for screening and prioritization of toxicants, including immunotoxicants. This review will survey the history, organizations, and projects that have pushed and continue to push the development of alternative methods, and it will provide examples of ongoing successful projects related to immunotoxicity.

The Move to Alternative Methods

Concern for the welfare of animals used in basic and applied research is not a recent development. Charles Hume introduced the concept of the 3 Rs—replacement, reduction, and refinement—in 1954 (Zurlo et al. 1996). A variety of regulations and projects, enacted or sponsored by public and private institutions, now exist that are dedicated to the development and application of alternative testing methods, and even a cursory inclusive review of these efforts is beyond the scope of this paper. A few of the events and organizations are mentioned below to provide perspective on efforts that have a direct impact on immunotoxicity testing.

Public debate on the need to test personal care products in laboratory animals culminated in the European Commission's Cosmetics Directive 76/768/EEC, originally published in 1976 (European Commission, 1976) and subsequently amended many times, which prohibits the sale of cosmetics tested in animals in European Union (EU) member states by 2013. Significant progress has been made in the development of alternative animal and non-animal testing methods that reduce or replace traditional tests. However, validation of complex non-animal methods for immunotoxicity (allergic sensitization), reproductive toxicity, carcinogenesis, and repeated-dose toxicity end points are unlikely to be in place by 2013; in 2009, a panel of experts estimated that a more realistic timeframe is 2017 to 2019 (Adler et al. 2011).

In 2001, the European Commission published a white paper (European Commission, 2001) that outlined future policy designed to protect the public health and the environment from adverse effects of new and existing chemical substances produced or used in the EU. The white paper is the precursor of the European Community Regulation, “Registration, Evaluation, Authorisation and Restriction of Chemical Substances that became law in 2007 (REACH, 2007). The legislation requires evaluation of chemicals used or produced in quantities exceeding one ton per year (an estimated 30,000 substances) for potential toxicity. Although the law does not specifically ban toxicity testing in animals, it does encourage development of models that reduce or replace the use of animals. Given that little or no toxicity data are available for the majority of the regulated substances, compliance will greatly increase the number of animals required to assess toxicity using traditional testing methods and take many years (decades) to complete. From a purely practical standpoint, the legislation provides a strong impetus to develop rapid and reliable alternative testing methods.

The need to develop alternative methods inspired programs in Europe and the United States that promote alternative methods and actively participate in validation of these methods. The European Centre for the Validation of Alternative Methods (ECVAM; http://ecvam.jrc.ec.europa.eu/) was created in 1991 in response to the Council of European Communities Directive 86/609/EEC, which requires members of the European Commission to support the development and validation of alternatives to animal testing in accord with the principle of the 3 Rs. A similar center, known as the Interagency Coordinating Committee on the Validation of Alternative Methods (ICCVAM; http://iccvam.niehs.nih.gov/), which began in 1994, was developed by the U.S. National Institutes of Health/National Institute of Environmental Health Services (NIEHS) National Toxicology Program (NTP) and is administered by the NTP Interagency Center for the Evaluation of Alternative Toxicological Methods. As their names imply, both organizations actively support the development and validation of alternative testing methods that reduce the number of animals required for safety testing or that replace in vivo models with in vitro alternatives.

In 2007, the U.S. National Research Council published Toxicity Testing in the 21st Century: A Vision and Strategy. The report was commissioned by the EPA in an effort to provide a framework for a comprehensive remodeling of toxicity testing practices to reduce animal use and increase testing capacity by exploiting developments in in vitro methods, computational models, systems biology, and in silico modeling to identify toxicity pathways and estimate potential exposures. These methods would be applied in a high-throughput format to screen chemicals for potential toxicity and prioritize flagged chemicals for targeted testing to improve risk assessment.

In parallel, EPA launched the National Center for Computational Toxicology, which directs the Agency’s Computational Toxicology research effort known as ToxCast (http://www.epa.gov/ncct/toxcast), launched in 2007. The ToxCast suite of assays includes 500+ high-throughput in vitro models based on primary and continuous cell lines to assess changes in gene and cell surface marker expression and nuclear receptor activation. Phase I of the project tested the “ToxCast 320” library of chemicals, primarily pesticide active ingredients, as a proof of concept exercise to determine toxicity signatures of the chemicals and concordance of the in vitro results with the extensive in vivo toxicity profile for the same set of chemicals. When this review was prepared, ToxCast data analysis did not specifically address immunotoxicity as an end point. However, the BioMAP set of assays included primary human endothelial and bronchial epithelial cells, fibroblasts, keratinocytes, peripheral blood mononuclear cells, and vascular smooth muscle cells that are cultured in the presence of stimulatory ligands relevant to immune system signaling (http://www.epa.gov/ncct/download_files/chemical_prioritization/BioSeek_ToxCast_Summary_24Jan08.pdf). These cells (with the exception of muscle cells) participate in the response to irritants and allergens and, as members of the innate and adaptive arms of the immune system, produce homeostatic and pathological cytokines. When used to assess the ToxCast 320 library, four clusters of toxic effects were identified: inhibition of mitochondrial function, inhibition of NFκB signaling, elevation of cAMP, and endoplasmic reticulum stress (Houck et al. 2009). All of these effects suggest potential immunotoxicity, and further data analysis to compare these effects to those obtained in traditional immunotoxicity testing would be interesting. Phase II of the project will test 1,000+ compounds including industrial chemicals, consumer products, food additives, and failed drug candidates, broadening the scope of potential modes of action. The NCCT effort includes the development of sophisticated modeling software, systems biology models to identify toxicity pathways, and relational databases that will eventually enable correlation of high-throughput data with effects observed in animals. The cause of high-throughput screening was recently expanded by a memorandum of understanding between the NTP, EPA, and the U.S. Food and Drug Administration, with the goal of testing 10,000 chemicals over the coming years.

Alternative Screening Methods for Immunosuppression

Ideally, tests to detect immunosuppression provide a direct measure of immune function that is mechanistically linked to maintaining homeostasis. An obvious example is vaccination and subsequent measurement of antibody titers. The response provides a holistic summation of antigen recognition, processing and presentation, gene transcription and rearrangement, cell proliferation and differentiation, and ultimately, release and distribution of effector molecules. Because of the many steps and variety of cells involved in protecting the host, and the existence of alternative pathways that may also offer protection, in vitro screening methods based on a few selected steps in the process may or may not reliably detect adverse effects, even if the models are well documented and critical to immunological success. However, a number of models that have been described and tested assess the supply of cells and markers of reduced or modified cellular function, including changes in gene expression, mediator (cytokine) production, or surface phenotype indicative of cell maturation or function.

Suppression of the Supply of Immunocompetent Cells

Pluripotent hematopoietic stem cells (HSC) mature in the bone marrow under the influence of growth factors produced by stromal cells. Stem cells then differentiate into precursors of myeloid cells that take part in innate immune responses and lymphoid cells, which mature in lymphoid organs into naïve thymus-derived T lymphocytes, under the influence of thymic epithelial cells, and bursal equivalent B lymphocytes. A variety of environmental chemicals have been shown to adversely affect stromal and thymic epithelial cells, decreasing the supply of progenitors and naïve T cells. Genarri et al. (2005) proposed an in vitro tiered approach to screen chemicals for adverse effects on the supply of cells, beginning with evidence of bone marrow toxicity; a positive finding would indicate a likely immunotoxicant based on disruption of cell supply, although a negative finding would not necessarily be conclusive, because some chemicals spare bone marrow but directly affect immune system cells. The second tier would assess lymphocyte viability to estimate effects on the supply of naïve lymphocytes. Reduced lymphocyte viability might then trigger additional testing using in vitro correlates of immune function, including natural killer cell activity, cytokine production, proliferation, and responses to alloantigens.

Assays to evaluate effects on HSCs have been used for years to investigate modes of immunotoxicant action. Hematopoietic stem cells are typically cultured in semisolid agar containing lineage-specific growth factors, and resulting colonies of cells are stained and counted. Recent assays have attempted to reduce the level of technical expertise required for traditional methods by assessing viability of the developing cells. Evaluation of a commercial bone marrow toxicity assay is in progress, under the direction of the NTP, to assess the effects of environmental chemicals that are known bone marrow toxicants, chemicals that tested negative in in vivo immunotoxicity tests, and other chemicals that have not been assessed for immunotoxicity (Germolec, personal communication). The assay assesses the viability of colony-forming cells from human and rodent HSCs that give rise to erythrocytes, granulocytes, and T and B lymphocytes, in a high-throughput format, thus preserving the ability to identify lineage-specific effects in a shorter period of time. Data analysis is in progress as of this writing. Haglund et al. (2010) reported the use of a 384-well plate format viability assay, using cryopreserved CD34+ hematopoietic progenitor cells isolated from human cord blood, to detect inhibition of cell growth (flow cytometric detection of viability) in the presence of myelotoxic drugs. They compared IC₅₀ values with reported values obtained in a traditional colony forming assay and found that values in the microculture system were similar. Even though culture times are similar to traditional colony-forming assays (seven days), assessing viability rather than staining and scoring colonies holds promise as a means to increase throughput.

Markers of Reduced or Modified Cell Function

Continuous rodent and human cell lines are readily available and preclude the use of living donors, although concerns have been raised that these malignant cells are not the physiological equivalents of primary cells. Cell lines do lend themselves to the development of transgenic lines expressing fluorescent reporter gene constructs that can be used to rapidly screen for changes in gene expression. One such reporter gene model is based on the rodent EL-4 thymoma cell line transfected with promoters for interleukin-2 (IL-2), interferon-γ (IFN-γ), IL-4, and IL-10 (Ullerås et al. 2005). The genes were selected to reflect T cell growth (IL-2), T helper 1 cytokine production (IFN-γ), T helper 2 cytokine production (IL-4), and inhibition of T helper 1 cytokine production (IL-10). Cells were cultured with and without ionomycin and phorbol to activate the cells in the presence of a variety of immunosuppressive drugs and environmental chemicals, as well as chemicals associated with increased allergy and autoimmunity (Ringerike et al. 2005). Twelve of the thirteen chemicals were reported to alter gene transcription, and the authors concluded that the model was best at detecting potential immunosuppression. In some cases, effects on transcription occurred only at a dose that decreased viability, and not all chemicals were tested at all doses, but the results suggest that their approach may be suitable for screening chemicals for immunosuppressive potential. In counterpoint, Koeper and Vohr (2009) monitored proliferation and cytokine production by freshly isolated rat spleen cell stimulated with T and B cell mitogens (concanavalin A and lipopolysaccharide), and in vitro antibody production to sheep erythrocytes by freshly isolated mouse splenocytes in the presence of some of the same compounds tested by Ringerike at al. (2005).

With the exception of cyclophosphamide, a cytotoxic drug that requires metabolic activation, the antibody response correctly identified immunosuppressive and nonimmunosuppressive chemicals. The proliferation and cytokine production assays were less predictive as a result of conflicting results with the two methods used to monitor viability or outright failure to detect immunosuppressive chemicals. The authors concluded that function (antibody production), rather than functional correlates, were necessary to correctly detect immunotoxicants. Although most immunotoxicologists would probably agree that functional assays are superior, if the goal is to flag potential immunotoxicants, reporter gene assays may prove to be useful. In keeping with the theme of assessing function, a recent publication by Collinge et al. (2010) described the HuLA (human lymphocyte activation) assay that used cryopreserved peripheral blood mononuclear cells isolated from humans after immunization with a standard influenza vaccine. Cells were cultured with dilution of the same vaccine preparation and assessed for antigen specific antibody production and proliferation. The assay detected dose-dependent suppression of both end points by immunosuppressive drugs with different modes of action and produced rankings of immunosuppressive potency based on IC₅₀ values. Furthermore, suppression was detected at therapeutic drug concentrations. The clear advantage of this approach is the use of human cells from individual donors to measure a highly relevant function. However, the test compounds were potent therapeutic agents used to deliberately suppress immune function; whether or not the method can detect unintended mild to moderate suppression caused by environmental chemicals is unknown.

Alternative Screening Methods for Allergic Hypersensitivity

Allergic hypersensitivity may be expressed as life-threatening anaphylaxis, allergic lung disease including asthma, food intolerance, or cutaneous reactions following dermal exposure. Significant research effort has been dedicated to understanding the pathophysiology and susceptibility factors that contribute to these types of maladaptive immune responses to antigens that pose no danger to the host. Allergic contact dermatitis (ACD) is a relatively common form of hypersensitivity in the general population; offending substances include plant products (e.g., poison ivy), metals, drugs, and industrial chemicals, particularly low molecular weight (<250 daltons) compounds that are too small to incite an immune response in their native state. Immunologists refer to this type of chemical as a hapten. Induction of contact sensitization requires that the sensitizer (hapten) penetrate the skin and bind to host proteins or peptides, forming an immunogenic hapten–protein complex. In some cases, the offending chemical is a prohapten that requires metabolic or photolytic conversion to an active protein binding state. Exposure must induce cytokine production by keratinocytes, resulting in local nonspecific inflammation, which in turn stimulates professional antigen presenting cells (in this case, Langerhans cells, a type of dendritic cell in the epidermis) to recognize the hapten, undergo maturation, and migrate to regional lymph nodes. There, hapten is presented to naïve T cells, which mature to become cytotoxic T cells, under the influence of cytokines produced by Langerhans cells; subsequent dermal exposure triggers migration of cytotoxic T cells to the point of contact and destruction of host cells bearing newly bound hapten. These well-documented steps in the sensitization process have inspired models that focus on chemical properties and on markers of cellular activation and proliferation.

The potential for a chemical to cause allergic sensitization is often related to its structure, although exposure conditions, dose per unit area, and other factors determine the outcome of human exposure. As noted in the introduction, regulatory agencies often mandate testing chemicals for potential sensitizing activity, and it is in this area of immunotoxicity screening that the development of alternative methods has made the most progress. Unlike immunosuppression, which may have many modes of action and go undetected unless suppression is sufficiently severe to increase the frequency and duration of infections, the events leading to ACD are relatively straightforward, and symptoms are usually obvious and easily recognized. One notable effort dedicated to alternative immunotoxicity testing methods is the Sens-it-iv project, a cooperative venture between industry, government, and academia created specifically for this purpose. The project developed training sets of sensitizers and negative controls and participated in the development and prevalidation of assay methods. The project is slated to end in November 2011 with a congress to review progress and discuss applications of the scientific achievements of the project.

Chemical Properties

Gerberick et al. (2008) reported the outcome of an ECVAM workshop on using chemical reactivity data to identify skin sensitizers. The report is comprehensive and an excellent resource of detailed information on the history, basic science, and regulatory environment related to ACD and alternative screening methods for chemicals that may cause ACD. The report is recommended as a source for greater detail than provided in this review.

Cellular activation: Keratinocytes.

These cells account for approximately 95% of the epidermis and respond to irritation and damage by producing proinflammatory cytokines and chemotactic factors. Van Och et al. (2005) evaluated cytokine production in keratinocyte cell lines derived from mice (HEL-30) and humans (HaCaT) to determine if culture for 24 hours in the presence of known moderate and strong contact sensitizers stimulated a unique cytokine production signature. Potency rankings based on intracellular concentrations of IL-1α and IL-18 in HEL-30 cells were similar to those derived from the mouse local lymph node assay but results from the human cell line were not as promising. Although the strongest sensitizer could be ranked as more potent than others tested, the results were disappointing from a risk characterization or management viewpoint because far less of a potent sensitizer is required to result in long-lasting hypersensitivity. A similar assay that assessed only intracellular IL-18 concentrations was described by Corsini et al. (2009) using the human keratinocyte line NCTC2455 cultured for 24 hours with a different set of chemicals that included known contact sensitizers, prohaptens, respiratory sensitizers and irritants. All contact sensitizers, including the prohaptens, increased IL-18 by at least 3 fold compared to non-exposed control cultures, whereas respiratory sensitizers and irritants did not. Primary human keratinocytes were also tested by this group against a limited subset of the test articles; similar results were obtained, suggesting that the cell line was an adequate substitute for primary cells and offers the technical advantages of using cultured cell lines versus collecting and growing primary cells. Success in the Corsini et al. (2009) human cell line model versus the disappointing results of Van Och et al. (2005) in another human line may reflect inherent differences in the cell lines or the extent of chemicals tested. Washing and lysing cells to release intracellular IL-18 after exposure was used in both assays to avoid potential false positive signals because of nonspecific release of IL-18 caused by cell damage. The twenty-four-hour culture period and washing step prior to IL-18 analysis by enzyme-linked immunosorbent assay suggest that these assays are not ideal for high-throughput screening.

Cellular Activation: Dendritic Cells

Langerhans cells are specialized skin DC. Numerous attempts have been made to develop alternative methods based on cytokine and chemokine production and changes in cell surface marker expression that coincides with cellular maturation of primary human DC and DC-like cell lines. A comprehensive review by Galvão dos Santos et al. (2009) concluded that no individual model currently available is able to differentiate contact sensitizers from irritants or to rank the potency of sensitizers. However, Mitjans et al. (2010) did report that model allergens, including a prohapten, upregulated p38 mitogen–activated protein kinase in the human THP-1 cell line (a monocytic leukemia cell that can be stimulated to develop DC-like attributes) after fifteen minutes of incubation. Release of IL-8 by THP-1 cells was also increased by all contact allergens except the prohapten isoeugenol, perhaps because of decreased stability of IL-8 mRNA.

Cellular Activation: Co-culture of DCs and Naïve T Cells

Martin et al. (2010) reported the results of a workshop on the use of T cell models to detect potential allergenicity of drugs, proteins, and chemicals. A primary focus of the meeting was the use of co-culture techniques to mimic and detect antigen presentation to and activation of naïve T cells, the penultimate event in contact sensitization. Test substance can be added directly to T cells, bound to protein before addition, or used to pulse DCs prior to co-culture, depending on the question to be answered. Multiple options are available to assess T cell activation, including up-regulation of cell surface markers indicative of antigen activation, production of cytokines, and cell proliferation. The co-culture models have great flexibility and appear to be ideally suited for investigating mode of action, identifying cellular targets, and generating primary human cell data for comparison with results from candidate human and rodent cell line models. In their present form, these models are unlikely to be high throughput but are important in the development of alternative methods and reducing animal use.

ECVAM Validation of Allergic Contact Hypersensitivity Models

Three screening assays were in the process of ECVAM prevalidation in June of 2011 (Aeby et al. 2010). A cell-free chemical test, known as the direct peptide reactivity assay, determines whether a test article satisfies the protein binding criterion for skin sensitizers. To assess protein binding in the direct peptide reactivity assay, two peptides were synthesized, containing either a single cysteine or lysine residue, favored targets for electrophilic attack by low-molecular-weight sensitizers (Gerberick et al. 2008). Binding of suspect sensitizers was monitored by high-performance liquid chromatography. Accuracy of the method was calculated to be 86% for a suite of known chemicals balanced for extreme, moderate, and weak sensitizers and non-sensitizers, based on LLNA testing. The other two models are based on separate human cell lines and mimic DC activation, a prerequisite for successful antigen presentation to T cells; these models are referred to as the Human Cell Line Activation Test (h-CLAT; Sakagguchi et al. 2010) and the Myeloid U937 Skin Sensitization Test (MUSST). In the h-CLAT assay, flow cytometry is used to detect expression of two surface markers (CD86 and CD54) of DC activation by THP-1 cells cultured for twenty-four hours with multiple dilutions of a test article. The assay was tested against a panel of chemicals that had tested positive or negative in the LLNA or were known positive or negative sensitizers in humans. Accuracy was calculated to be 84% versus LLNA results and 80% versus human data. Specificity values versus the LLNA and human data were 75% and 69%, respectively, indicating good accuracy at the price of somewhat increased false negatives and positive results. This assay was vetted in a series of “ring tests” in multiple laboratories to establish reliability and reproducibility among labs (Sakagguchi et al. 2010). The MUSST protocol is similar, although the incubation time was forty-eight hours for the U397 cell line, and only CD86 expression was assessed. Accuracy was calculated to be 83%, and specificity to be 84% versus human data for ninety-nine chemicals. If accepted by ECVAM, these testing methods will provide an alternative to traditional in vivo testing methods.

Summary

Efforts to develop alternative, non-animal methods to screen and rank the sensitizing potential of chemicals have had greater success than those to screen for immunosuppressive chemicals, aided by our understanding of the defined immunological events that must occur to induce sensitization. However, alternative immunosuppression screening and prioritization methods have been published that generate biologically plausible signals after exposure to a variety of environmental chemicals and drugs. Success may depend on defining the circumstances under which monocultures of cell lines return a signal that can be interpreted as an indicator of immunosuppression. Single-cell models lack cell–cell interactions critical to generating an immune response, do not reflect known alternative pathways for destroying or neutralizing infectious agents, and lack input from hormones and neurotransmitters that play critical roles in immune system development and homeostasis.